The present invention relates generally to a method for correcting defects formed on a photomask, more specifically, to a method for correcting defects formed on a alternating phase shift photomask.
The attempt to increase integration density of semiconductor circuits has resulted in the significant size reduction of patterns used to form the semiconductor circuits. Photomasks are utilized by the semiconductor circuit industry to produce the minute patterns onto semiconductor materials or wafers through photolithographic processing during the manufacturing of these integrated semiconductor circuits.
FIG. 1 depicts a sectional view of a portion of a standard photomask 52. Referring to FIG. 1, a typical mask has a metal light shielding film or layer 54 of a prescribed pattern formed on a mask substrate 56. When the metal light shielding film 54 is formed through known techniques an opaque defect (excess metal film) 72 may be generated. FIGS. 2-4 illustrate one example the generation of an opaque defect 72 on a mask 52 in the manufacturing process of a standard mask 52. FIG. 2 depicts one of the initial stages in the generation of a mask 52. A layer of light shielding film 54 is disposed over the entire surface of mask substrate 56. Mask substrate is normally formed of a transparent silica, quartz or other material well known in the art. Light shielding film 54 is typically formed of elements and compounds consisting of Cr, Mo, F, Si, Zr, O, N. A layer of photo sensitive polymer resist 58 is deposited over the entire surface of layer of light shielding film 54. After resist 58 is coated over the metal light shielding layer 54, a prescribed portion of resist 58 is exposed to a radiation source 64. If foreign material 66 is present in or on resist 58 at the time of exposure, in a region of mask 52 to be exposed, the portion of resist positioned under foreign material 66 will not be exposed resulting in a mask defect. Referring to FIG. 3, if the region to be originally exposed is thus not exposed due to foreign material 66, unnecessary resist artifacts 68 are formed when resist 58 is developed. Referring to FIG. 4, when unnecessary resist artifact 68 is formed, and the metal light shielding film 54 is subjected to etching, utilizing resist 58 as an etching pattern, a metal light shielding film opaque defect 72 is formed. Alternatively, due to incorrect exposure or an error in the design pattern, other areas of resist 58 also may not be exposed resulting in similar defects in mask 52. FIG. 5 depicts a standard method to correct opaque defects 72 in metal light shielding film 54 utilizing an Nd:YAG laser beam or alternatively a focus ion beam (FIB) 74. FIB 74 is irradiated onto opaque defect 72. Thus, opaque defect 72 is evaporated or sputtered and removed, as shown in FIG. 6. An Nd:YAG laser beam typically used has a wavelength of about xcx9c532 nm and irradiated for about 200 mJ-300 mJ, while a typical FIB 74 uses a dose of approximately 400-600 xcexcJ.
The prior art methods for correcting the opaque defect 72 of metal light shielding material have provided satisfactory results in correcting opaque defects 72 in standard masks. However, when using a laser to remove the metal light shielding material, the metal light shielding material is not removed, but is displaced or redeposited in the surrounding areas causing material swelling in the mask. Additionally, the laser spot size is large, thus, limiting the ability to repair highly integrated patterns. Further, the use of lasers tends to remove a portion of the substrate under the defect depending on the defect size. Because of the heat required to ablate the opaque defect 72 of light shielding material, the underlying substrate 56 is usually partially ablated or melted through thermal conduction, thus, reducing the effectiveness and precision of mask 52.
The effectiveness of an FIB 74 in the removal of opaque defect 72 is, to a large extent, based on the fact that a contrast distinction exists between the defect 72 and the transmitting window 78. Because FIB devices utilize secondary charged particles for imaging (i.e. secondary electrons or ions), a contrast between the defect and the correctly etched areas must be present to properly determine the size of the defect and the amount of light shielding material to be removed. With a standard mask 52, the contrast is large, the defect consists of a metal light shielding material while the correctly etched area is a transparent material. Thus, an FIB device can determine the size and amount of correction. However, without a significant contrast, an FIB device would not be able to accurately determine the defect size nor the degree of correction needed to accurately correct the defect or to avoid further damage to the surrounding mask.
The need to increase integration density has forced the positioning of the light transmitting windows or regions 76a and 76b closer together, as shown in FIG. 7. However, light transmitted through transmitting windows 76 will overlap if transmitting windows are positioned too close together. The light intensity 78 of the light transmitted through transmitting windows 76a and 76b is shown in FIG. 8. Light intensity 78a transmitted through window 76a will overlap light intensity 78b transmitted through window 76b. Thus, a total light intensity 80 results producing an in inaccurate integrated circuit.
In order to achieve further increased integration density a photomasking technique has been developed which provides for an alternating phase shift in the light which is projected through the mask. This type of photomasking is known as alternating phase shift photomasking or Levenson phase shift masking.
FIG. 9 depicts a sectional view of a portion of an alternating phase shift photomask or Levenson phase shift mask 120. Phase shift mask 120 includes a transparent silica substrate 122, including a phase shift region or layer 124 formed in substrate 122, and a radiation shielding film 126. For simplicity, this specification assumes the radiation used to form an integrated circuit pattern is light, although other types of radiation are also suitable. When radiation, for example from a light source, is shown or exposed onto mask 120, light passes through the regions of mask 120 which are not covered by light shielding film 126. The light which passes through mask 120 will be affected by the phase shift layer 124. Because of differing heights of phase shift layer 124, the light passing through it will exit mask 120 with different phases. The transparent regions or phase shift areas 130a and 130b are specifically arranged in an alternating pattern such that a first phase shift layer thickness (for example at 130a) is positioned next to a second phase shift layer thickness (130b) such that light passing through transparent regions 130a and 130b will exit at alternating phases thus producing a cancelling effect. The height differences of the phase shift layer 124 are specifically calculated according to the exposure lambda (xcex) and index of refraction (n) of the substrate step depth (d), defined by: d="PHgr"/xcfx80xcex/(nxe2x88x921).
In FIG. 10, the electric field of the transmitted light passing through mask 120 is shown. The light transmitted through region 130a of mask 120 produces a negative electric field 142 due to the lack of phase shift layer 124. The light transmitted through region 130b of mask 120, and thus phase shift layer 124, results in a positive electric field 144 and is thus 180xc2x0 out of phase with the light transmitted through region 130a. 
Due to the placement of the alternating transparent regions, the light transmitting through transparent regions 130a and 130b will produce a cancelling or destructive interference on those regions between the transparent regions 130a and 130b. Thus the exposure resolution and depth is greatly improved. FIG. 11 depicts the resulting transmitted light intensity of the light being transmitted through mask 120. Because of the cancelling effects of the alternate phases, the light intensity is ideally at a zero level 148 between transparent regions 130a and 130b, and at a maximum positive level 150a and 150b at the transparent regions 130a and 130b. 
When the substrate etching process to form a phase shift photomask is performed, any foreign material in the photo resist or on the substrate will produce a defect in the mask. The foreign material will interfere with the etching process thus producing an unetched bump or incomplete etching in phase shift layer 124. These defects are three-dimensional in nature. Use of a laser beam cannot remove the phase shift defect, because the phase shift defect is formed of a transparent material like SiO2, and a laser beam will transmit directly through the phase shift defect, thus having no effect on the phase shift defect. Further, a laser would not provide the accuracy needed because the laser spot size is so large in comparison to the size of both the mask details and defects.
Defects within a mask make the mask virtually useless, thus drastically increasing the costs to manufacture integrated circuits. Due to the extreme accuracy needed in the manufacture of Levenson phase shift masks and the size of the defects created, simply directing an FIB onto a phase shift defect cannot accurately correct the defects. Because the defect and the substrate are the same material, there is no contrast to distinguish the defect from the substrate. Thus, the FIB cannot accurately determine the size of the defect nor the amount of correction needed. Further, because the FIB device cannot determine size, location and amount of correction needed, the FIB can cause extensive damage to the substrate or surrounding phase shift material. Therefore, an effective and accurate method of correcting phase shift defects is needed.
A method and apparatus for correcting phase shift defects in a photomask is provided by scanning the photomask for the defect and determining locations of at least one defect. Following the determination of the location of a defect, the defect is three-dimensionally analyzed producing three-dimensional results. Utilizing the three-dimensional results, a focus ion beam (FIB) is directed onto the defect to eliminate the defect. The FIB is controlled by an etch map which is generated based on the three-dimensional results. To provide further precision to the repairing of the photomask, test patterns of the FIB are generated and three-dimensionally analyzed. The three-dimensional test pattern results are further utilized in generating the etch map to provide greater control to the FIB.
The method and apparatus of the present invention provide for the precise removal of residual phase shift defects. Because the defects are three-dimensional, simply directing an FIB at the defect will not accurately remove the defect. By providing a three-dimensional analysis of the defect, the present invention provides for greater precision in correcting the defect and thus produces a much more highly accurate alternating phase shift photomask. Further, by providing for the analysis of test patterns of the beam being utilized to correct the defect, even greater precision is obtained.